FILTERING HORIZON- SENSOR MEASUREMENTS FOR ORBITAL NAVIGATION
ROBERT FITZGERALD · Guidance and Control Conference · 1966
Optimal filtering theory is applied to the problem of processing horizon-sensor data so as to compensate for the unusual statistical properties that characterize the associated errors. An examination of the statistical characteristics of the variations in apparent-horizon altitude, together with the geometry of the measurement process, reveals correlation functions of unusual form corresponding to the various error components present in the horizon-sensor measurements. Filtering techniques are then derived by the application of optimal filtering theory to approximate statistical models of varying degrees of complexity. An essential operation is the fitting of simple correlation-function models to the true correlation functions exhibited by the random processes in question. It follows that proper choice of the measurement schedule can greatly enhance the effectiveness of the filter employed. The random functions involved are basically space-dependent rather than time-dependent; two methods are therefore given for the generation of a correlated random process on the surface of a sphere, for use in Monte Carlo simulations of the variations of horizon altitude. Numerical results are presented to demonstrate the effectiveness of the techniques developed.